Transcript G060506-00

Optical Coating Development for the Advanced
LIGO Gravitational Wave Antennas
Gregory Harry
LIGO/MIT
- On Behalf of the Coating Working Group University of Sannio at Benevento
October 9, 2006
Benevento, Italy
LIGO-G060506-00-R
Gravitational Wave
Detection
• Gravitational waves predicted by Einstein
• Accelerating masses create ripples in space-time
• Need astronomical sized masses moving near
speed of light to get detectable effect
LIGO
End Test
Mirror
Laser Interferometer Gravitational-wave Observatory
• Two 4 km and one 2 km long interferometers
• Two sites in the US, Louisiana and Washington
• Similar experiments in Italy, Germany, Japan
• Whole optical path enclosed in vacuum
• Sensitive to strains around 10-21
Whole Interferometer
Enclosed in Vacuum
Input Test
Mirror
Recycling
Mirror
4 km Fabry-Perot cavities
100 W
Laser/MC
6W
0.2 W
13 kW
2
LIGO Sensitivity
Seismic noise < 40 Hz
Measured sensitivity 6/2006
Optics sit on multi-stage
vibration isolation
Laser shot noise > 200 Hz
10 W frequency and
amplitude stabilized laser
Thermal noise 40 Hz < f < 200 Hz
Metal wire pendulum suspensions
allow optic to move freely with gravity
wave
3
Current LIGO Noise
Present noise at design value
in all three interferometers
 Some excess noise < 50 Hz
 Noise reduction during breaks
Currently taking data
 Will collect 1 years worth of
triple coincidence
 Began in November 2005
 Extensive data analysis ongoing





Hanford 4 K sensitivity
Neutron star inspirals 14.5 Mpc
10 MO black hole inspirals to 50 Mpc
Stochastic background 7.5 10-6
Crab pulsar e 2.8 10-5
Sco X-1 e 3.0 10-7
4
Advanced LIGO
Advanced Configuration
power recycling
mirror
signal recycling
mirror
Suspension
Improved
coating
silica
ribbon
Proposed Sensitivity
• Factor of 15 in strain improvement
• Seismic isolation down to 10 Hz
• 180 W of laser power
• Larger optics with improved coating
• Additional mirror for signal recycling
5
Advanced LIGO
Sensitivity
Initial LIGO Coating – Tantala/Silica
Limits sensitivity 40 Hz – 400 Hz
Thermooptic Noise high in same BW
Need improved coating – including
Brownian thermal noise, coating
thermoelastic noise, and coating
thermorefractive noise
Brownian thermal noise limits at low
frequency, even with reduced laser
power/radiation pressure noise
Thermal noise also limits narrowband
sensitivity, sets floor
Initial LIGO Coating
Binary Neutron Star Inspiral
160 Mpc
Binary Black Hole Inspiral
910 Mpc
Neutron Star/Black Hole Inspiral 360 Mpc
Stochastic Background
1.3 10-9
6
Measurement
Techniques
Coating Thermal Noise
• Q measuring on coated disks
• Can test many candidate coatings
•Thin – low freq – MIT, HWS, ERAU
•Thick – high freq - Glasgow
•Cantilevers – very low freq – LMA, Glasgow
Thin Sample
• Direct thermal noise
measurements at the TNI
TNI Result of Tantala/Silica Coating
Thick Sample
PCPI
Setup
Optical Performance
• Absorption measurements using
photothermal common path
interferometry (Stanford, LMA)
• Developments with initial LIGO optics
• High Scatter
• High Absorption
7
Initial LIGO
Tantala/Silica Coating
Coating Mechanical Loss
Layers
Materials
Loss Angle
al/4 SiO - l/4 Ta O
30
2.7 10-4
2
2 5
al/8 SiO - l/8 Ta O
60
2.7 10-4
2
2 5
al/4 SiO – l/4 Ta O
2
2.7 10-4
2
2 5
al/8 SiO – 3l/8 Ta O
-4
30
2
2 5 3.8 10
a3l/8 SiO – l/8 Ta O
-4
30
2
2 5 1.7 10
30
30
30
a LMA/Virgo,
bl/4
SiO2 – l/4 Ta2O5
cl/4 SiO – l/4 Ta O
2
2 5
dl/4 SiO – l/4 Ta O
2
2 5
3.1 10-4
4.1 10-4
5.2 10-4
Lyon, France
b MLD Technologies, Mountain View, CA
c CSIRO Telecommunications and Industrial Physics, Sydney, Australia
d Research-Electro Optics, Boulder, CO
Tantala/Silica Coating Mechanical Loss
Direct Coating Thermal Noise Measurement
No effect from from interfaces between
layers nor substrate-coating
Internal friction of materials seems to
dominate, with tantala having higher
mechanical loss
Noticeable differences between vendors
f – Ta2O5 (3.8 ± 0.2) 10-4 + f(1.1±0.5)10-9
f – SiO2 (1.0± 0.2) 10-4 + f(1.8±0.5)10-9
8
TiO2-doped Ta2O5
Examined titania as a dopant into
tantala to try to lower mechanical
loss
f1
f2
f3
f4
f5
=
=
=
=
=
(2.2±0.4)10-4
(1.6±0.1)10-4
(1.8±0.1)10-4
(1.8±0.2)10-4
(2.0±0.2)10-4
Titania-doped Tantala/Silica Coatings
+ f(1.2±0.6) 10-9
+ f(1.4±0.3) 10-9
+ f(-0.2±0.4)10-9
+ f(1.7±0.6) 10-9
+ f(0.1±0.4) 10-9
G. M. Harry et al, Submitted to Classical and
Quantum Gravity, gr-qc/0610004
TNI Noise from Titania doped Tantala/Silica
Young’s modulus and index of
refraction nearly unchanged
from undoped tantala
Optical absorption acceptable
≈ 0.5 ppm
9
Advanced LIGO Baseline
Coating
Baseline Advanced LIGO Coating –
Titania doped Tantala/Silica
Still not limited by quantum noise
Limits sensitivity 40 Hz – 200 Hz
Thermo-optic Noise high in same BW
Narrowband high frequency
configurations still limited by coating
thermal noise
Acceptable impedance match with
substrate
Acceptable coating thickness
Advanced LIGO Baseline TitaniaBinary Neutron Star Inspiral
doped Tantala/Silica
175 Mpc
Binary Black Hole Inspiral
975 Mpc
Neutron Star/Black Hole Inspiral 390 Mpc
Stochastic Background
1.2 10-9
10
Advanced LIGO Backup
Coating
Ratio(Si:Ti) Absorption Index
Run 1
50/50
1.5 ppm
2.15
Run 2
65/35
0.5 ppm
1.85
Y
87 GPa
73 GPa
Thick Sample – Run 1
f= (2.4 +/- 0.9) 10-4
Thin Sample
Run 1* f = (3.1 +/- 0.2) 10-4
Run 2 f = (1.9 +/- 0.3) 10-4
•
•
•
•
Silica doped Titania/Silica
-Backup Coating-
Low Young’s Modulus
Low Index (Thicker Coating)
Good Mechanical Loss
Good Optical Absorption
Binary Neutron Star Inspiral
175 Mpc
Binary Black Hole Inspiral
960 Mpc
Neutron Star/Black Hole Inspiral 385 Mpc
Stochastic Background
1.2 10-9
11
Other Coatings
Attempted
• Niobia/Silica – high mechanical loss, unknown optical absorption
• Hafnia/Silica – poor adhesion, poor absorption, never measured for f
• Alumina/Silica – thick coating, good mechanically and optically
• Dual ion beam (oxygen) – interesting, shows differences in mechanical
loss between masks but not improvement over baseline
• Oxygen poor – high mechanical loss, waiting on annealing in nitrogen
atmosphere, high absorption
• Xenon ion beam – increased mechanical loss
• Lutetium doped Tantala/Silica – no improvement in mechanical loss
• Differing annealings – inconclusive, no major improvements,
absorption issues
• Effect of substrate polishing – no effect on mechanical loss
• Most of these do not have Young’s modulus measurements or optical
absorption
12
New Coating Materials
Ozone annealing – improve stoichiometry
Neon ion beam – xenon made things worse
Alumina as dopant into Ta, Ti, or Si
Tungsten dopant into Ta (and Ti, Nb, Hf, etc)
Zirconia
Hafnia – solve adhesion problem
Cobalt as dopant – only layers near substrate
Dopants:high index-high index
Hf-Ta
Nb-Ti
Hf-Nb, etc
Trinary alloys
Ta-Ti-Si
Ni-Ta-Ti-Zr-Hf-Si-Al
Si-O-N
Other nitrides
13
Thermo-optic Noise
•Coating thermorefractive (b=dn/dT) and
coating thermoelastic noise (a=dL/dT) are
coherent noise sources
•Combined noise – Thermo-optic Noise
•Best number in literature indicates very
high thermorefractive noise from tantala
b = 1.2 10-4
•Thermo-optic noise at this level ruled out
by TNI upper limits
•Almost certainly wrong, but what is the
right value?
•Significant reduction in sensitivity
Advanced LIGO with High
Thermorefractive Noise
Binary Neutron Star Inspiral
150 Mpc
Binary Black Hole Inspiral
910 Mpc
Neutron Star/Black Hole Inspiral 340 Mpc
Stochastic Background
1.4 10-9
14
dn/dT Measurement
•Thermorefractive(b=dn/dT)/coating
thermoelastic noise(a=dL/dT) noise correlated
• b from literature (Inci J Phys D:Appl Phys, 37 (2004) 3151)
1.2 X 10-4
• This value makes combined noise an AdvLIGO
limiting noise source
• Limits from TNI encouraging that b is lower
• Need a good value for tantala, titania doped
tantala, and other promising coatings
• Experiment at Embry-Riddle
Aeronautical University
• Measure change in reflectivity
versus temperature
• Use green He-Ne laser at 45 degrees
• 100 C change in temperature enough
to verify/rule out Inci result for tantala
15
Young’s Modulus of
Coatings
Coating Young’s modulus just as important to thermal noise as
mechanical loss
Acoustic reflection technique used to measure coating impedance
in collaboration with Stanford (I Wygant)
MLD alumina/tantala 176 +/- 1.1 GPa
MLD alumina/tantala 167 +/- 1.3 GPa
MLD silica/tantala
91 +/- 7.0 GPa
WP alumina/tantala 156 +/- 20 GPa
Fit of Young’s Modulus of Tantala/Alumina
Uses assumed values for material densities
Infer material Young’s moduli
YTa2O5 = 140 +/- 30 GPa
YAl2O3 = 210 +/- 30 GPa (MLD)
YAl2O3 = 170 +/- 30 GPa (WP)
Large errors problematic when propagated
16
Study of Materials
X-Ray Florescence Results from Southern
University / CAMD
•Measurements being made at
Glasgow, Southern, and
Caltech
• Titania concentrations in
titania-doped tantala consistent
– LMA/SU/UG
• Southern finding titania using
XRF, XANES, EXAFS
• Plans for AFM and GIXAFS at
Southern
• Hopes for further insights into
coating makeup and structure
from studying contaminants
Electron Energy Loss
Spectroscopy results from
Glasgow
17
Modeling and Molecular
Cause of Mechanical Loss
Goal: A description of mechanical loss in thin film amorphous oxides
from basic principles
Molecular dynamics calculations beginning at University of
Florida
•Have a working semi-empirical model of loss in fused silica
•Frequency dependence from two level systems
•Surface loss as observed phenomenon
•Develop full molecular description of silica loss
•Surface loss caused by two member rings
•Generalize to other amorphous oxides
•Analogous two level systems
4.5E-04
Mechanical loss data at different
temperatures
•Tantala/Silica T>300 C
•Ti doped Tantala/Silica T<300 C
•With frequency dependence, start
to fit to modeling
Calculated coating loss
4.0E-04
3.5E-04
3.0E-04
2.5E-04
2.0E-04
1.5E-04
1.0E-04
0
50
100
150
200
Temperature (K)
250
300
350
18
Thermal Noise inThird
Generation
Crucial to improve beyond Advanced LIGO levels to exploit QND, very low
frequency seismic isolation, improved topologies, high laser power, etc
•Short cavities as reflectors
•Khalili (Phys Lett A 334 (2005) 67)
•Significant added complexity
•No experimental work so far
•Corner reflectors
•Braginsky and Vyatchanin (Phys Lett A 324 (2004) 345)
•Practical concerns (scatter, finess, angular stability, etc)
•Experiment at Australian National University -99.89% reflectivity observed
•Lower temperatures
•Need to restudy all materials as properties change
•Some preliminary experimental work
•New substrate materials (sapphire, silicon, etc)
•Will require new coatings
•Possibly dopants added to substrates
•Change in beam shapes
•Mesa beams – better averaging of thermal fluctuations
•Higher order modes
•General theory from O’Shaughnessy/Lovelace
•Experiments at Caltech
19
Thermal Lensing
•Absorption of optical power in mirrors
causes heating
•Most absorption in coatings because of
higher power in the Fabry-Perot cavities
•Heating of optic causes physical
distortions and changes in index of
refraction
•Optical path length changes distorts
wavefront
•Causes poor contrast defect, ultimately
increased shot noise and poor sensitivity
•Thermal lensing can be corrected by
adding heat to cold parts of optic
•Use ring heaters or CO2 lasers
•Limit to how much heat can be
provided
•Inhomogeneous absorption requires
scanning laser system
•Increase in rad pressure noise
•Complicated controls
•Need coatings to have absorption
≤ 0.5 ppm and homogeneous
20
Excess Absorption at
Hanford
• Input optics curved to match
recycling mirror curvature at 8 W
• Point design assumes a value
for absorption
• Found best matching at 2.5 W
• Additional absorption causes
excess thermal lensing
• Excess absorption has to be in
recycling cavity optic
• Input mirrors or beamsplitter
Other interferometers (2 K at
Hanford and 4 K at Livingston)
found to have much less
absorption than expected
Sideband Recycling Gain
LIGO 4K Hanford IFO
21
Initial LIGO Thermal
Compensation Design
• 8 W CO2 laser directly projected onto mirrors
• Ring heater not used to minimize installation time in vacuum
• Scanning laser not used to avoid Shack-Hartmann sensors and
radiation pressure issues
• Different masks used to compensate for high or low absorption
• Laser power controlled by acousto-optic modulator (H2) and
rotating polarization plate (H1, L1)
• Power controlled by feedback from IFO channels
22
Bench Tests of H1:ITMx
• H1:ITMx shipped to Caltech
immediately after removal
• Absorption measured using
photothermal common-path
interferometry
• Background < 1 ppm
• Significant outliers with
absorption > 40 ppm
Dust source of absorption?
• Soot from brush fire in 2000?
• Attracted by charged surface?
• Insufficient cleaning and
handling procedures?
23
Conclusions
•Coating thermal noise limiting noise source in Advance LIGO’s most
sensitive frequency band
•Determined source of coating mechanical loss is internal friction in
constituent materials
•High index, typically tantala, is the biggest source of thermal noise
•Doping a means of reducing mechanical loss
•Titania doped into tantala
•Silica doped in titania
•Many other techniques tried to improve thermal noise, many still
to be pursued
•Thermo-optic noise a potential problem that is understudied
•Need more information on coating Young’s moduli
•Much work to be done with characterizing coating materials and
developing thermal noise theory
•New ideas for third generation only beginning to get attention
•Absorption and scatter high in Initial LIGO
• Both at levels that would not be acceptable in Advanced LIGO 24
Theory
Sx(f) =d(1-s2)/(p w2)((1/(Yperp (1-s2))-2 s22Ypara/(Yperp2 (1-s2)(1-s1))) fperp+
Yparas2(1-2s)/(YperpY(1-s1)(1-s))(fpara-fperp)+Ypara(1+s)(1-2s)2/(Y2(1-s12)(1-s))fpara)
What we have
•Complete theory of infinite mirror from Levin’s theorem
•Anisotropic coatings including Young’s modulus, loss angles, and
Poisson ratios
•Relationship between total anisotropic coating parameters and
isotropic individual material parameters
•FEA models of finite mirror effects
•Theory of coating thermoelastic loss
•General theory of coatings and substrates, both Brownian and
thermoelastic, for any beam shape for infinite mirrors
•Optimization of coating thicknesses for thermal noise and reflectivity
(see talk by V Galdi)
What we need
•Empirical formula for finite mirror effects
•Analytical theory of finite mirrors
•Molecular level description of loss angles and other parameters
•Complete optimization over thermal noise, reflectivity, absorption,
scatter, etc.
25
Scatter in Initial LIGO
26
Excess Absorption at
Hanford
• Three techniques used to
determine source of excess
absorption
• Change in g factor
• Thermal compensation power
• Change in spot size
• Fairly consistent result (assuming
absorption in HR coating)
• ITMx 26 ppm
• IMTy 14 ppm
• Design 1 ppm
• Resulting changes
• ITMx replaced
• ITMy drag wiped in situ
Spot size measurements: Data and technique
27
Absorption improvement
at Hanford
• ITMx replaced with spare optic
• ITMy drag wiped in place
• Both optics (ITMx and ITMy)
show improved absorption
• Both < 3 ppm
• Power 6.8 W - mode cleaner
• Shot noise at design level
• 15 Mpc binary neutron star
inspiral range
28
Thermal Compensation
Upgrades and Challenges
•
•
•
Initial LIGO compensation effective at 100 mW absorbed
Advanced LIGO expected to have 350 mW absorbed
Cleanliness and handling will be crucial
» Need to keep absorption down
•
Potential improvements for advanced detectors
»
»
»
»
»
•
Graded absorption masks
Scanning laser system
Compensation plate in recycling cavity
Graded absorbing AR coating
DC readout, reducing requirements on RF sidebands
ITM Compensation
Plates
Challenges
»
»
»
»
»
Greater sensitivity
New materials – sapphire ~ 20 ppm/cm absorption
Compensation of arm cavities
Inhomogeneous absorption
Noise from CO2 laser
PR
M
ITM
SR
M
29
Advanced LIGO
Thermal Compensation
•
Ring heaters simplest compensation system
» Adds a lot of unnecessary heat
» Could cause thermal expansion of other parts
•
Scanning laser system causes noise
» Jumps in location cause step function changes in thermal expansion
» Harmonics of jump frequency could be in-band
» Could require feedback with Hartmann sensors or similar
•
Staring laser system works on Initial LIGO
» Could require unique masks for each optic
» Unique masks could be inappropriate as system is heating up
» CO2 laser noise still a problem
30